Accurate and efficient regression modeling for microarchitectural performance and power prediction

被引:0
作者
Lee, Benjamin C. [1 ]
Brooks, David M. [1 ]
机构
[1] Harvard Univ, Div Engn & Appl Sci, Cambridge, MA 02138 USA
关键词
microarchitecture; simulation; statistics; inference; regression;
D O I
10.1145/1168918.1168881
中图分类号
TP31 [计算机软件];
学科分类号
081202 ; 0835 ;
摘要
We propose regression modeling as an efficient approach for accurately predicting performance and power for various applications executing on any microprocessor configuration in a large micro-architectural design space. This paper addresses fundamental challenges in microarchitectural simulation cost by reducing the number of required simulations and using simulated results more effectively via statistical modeling and inference. Specifically, we derive and validate regression models for performance and power. Such models enable computationally efficient statistical inference, requiring the simulation of only 1 in 5 million points of a joint microarchitecture-application design space while achieving median error rates as low as 4.1 percent for performance and 4.3 percent for power. Although both models achieve similar accuracy, the sources of accuracy are strikingly different. We present optimizations for a baseline regression model to obtain ( 1) application-specific models to maximize accuracy in performance prediction and ( 2) regional power models leveraging only the most relevant samples from the microarchitectural design space to maximize accuracy in power prediction. Assessing sensitivity to the number of samples simulated for model formulation, we find fewer than 4,000 samples from a design space of approximately 22 billion points are sufficient. Collectively, our results suggest significant potential in accurate and efficient statistical inference for microarchitectural design space exploration via regression models.
引用
收藏
页码:185 / 194
页数:10
相关论文
共 15 条
[1]  
[Anonymous], INT C PAR ARCH COMP
[2]  
[Anonymous], 1986, STAT SCI
[3]   New methodology for early-stage, microarchitecture-level power-performance analysis of microprocessors [J].
Brooks, D ;
Bose, P ;
Srinivasan, V ;
Gschwind, MK ;
Emma, PG ;
Rosenfield, MG .
IBM JOURNAL OF RESEARCH AND DEVELOPMENT, 2003, 47 (5-6) :653-670
[4]  
EECKHOUT L, 2003, IEEE MICRO SEP
[5]  
Harrell FE Jr, 2001, REGRESSION MODELING
[6]  
IPEK E, 2006, ASPLOS, V12
[7]  
IYENGAR V, 1996, S HIGH PERF COMP ARC
[8]  
JOSEPH P, 2006, S HIGH PERF COMP ARC
[9]  
KARKHANIS T, 2004, INT S COMP ARCH JUN
[10]  
LEE B, 2006, TR0806 HARV U